Intelligent visual multifunctional full-automatic intestinal feeding instrument controlled by AI

Abstract

The invention relates to the technical field of medical monitoring equipment, in particular to an AI-controlled intelligent visual multifunctional full-automatic intestinal feeding instrument, which comprises a multi-mode sensing assembly, a control module, a signal processing module, a signal processing module and a signal processing module, and is characterized in that the multi-mode sensing assembly comprises a plurality of sensors for monitoring various intestinal information of a patient; the monitoring equipment host receives and displays the intestinal tract information; the feeding module is used for feeding the patient; the data analysis module is used for analyzing the abnormity of the intestinal information, comprehensively analyzing the intestinal condition of the patient by adopting the constructed time-space synchronization fusion model and combining different intestinal information, and generating an intestinal condition analysis result; correspondingly, the monitoring equipment host is triggered to perform display and early warning, and the feeding module is regulated and controlled to feed the patient. According to the scheme, multi-aspect accurate gastrointestinal tract monitoring can be carried out on a patient with an indwelling jejunum feeding tube, monitored data are mined and analyzed, feeding is automatically adjusted, the safety of the patient is guaranteed, and the rehabilitation speed is increased.

Description

Technical Field

[0001] This invention relates to the field of medical monitoring equipment technology, specifically to an AI-controlled intelligent visual multifunctional fully automatic enteral feeding device. Background Technology

[0002] Critically ill patients in the intensive care unit often require jejunal feeding tubes due to their inability to feed themselves. However, jejunal tubes have a limited function, making it impossible to visualize the internal condition and changes of the gastrointestinal tract or identify any lesions. Furthermore, the amount of nutrient solution provided is determined solely based on the patient's height, weight, and blood test results, lacking the integration of gastrointestinal monitoring data. This makes it difficult to achieve scientific nutrition, leading to various health problems and even life-threatening situations for many patients due to nutritional deficiencies.

[0003] To address the aforementioned issues, researchers have developed a gastrointestinal monitoring system. However, this system only has instruments capable of monitoring esophageal pH and pressure, which cannot provide real-time readings and lacks an imaging system. Furthermore, gastrointestinal testing relies primarily on gastroscopy and colonoscopy sampling, and the monitored data is displayed in a single format without further data analysis, thus limiting its widespread adoption and application.

[0004] Therefore, there is an urgent need for an AI-controlled intelligent visual multifunctional fully automatic enteral feeding device that can perform multi-dimensional and precise gastrointestinal monitoring for patients with indwelling jejunal feeding tubes, and analyze the monitoring data to automatically adjust feeding, ensure patient safety, and accelerate recovery. Summary of the Invention

[0005] The present invention aims to provide an AI-controlled intelligent visual multifunctional fully automatic enteral feeding device, which can perform multi-faceted and precise gastrointestinal monitoring for patients with indwelling jejunal feeding tubes, and mine and analyze the monitoring data to automatically adjust feeding, ensure patient safety, and accelerate recovery.

[0006] This invention provides the following basic solution: an AI-controlled intelligent visual multifunctional fully automatic intestinal feeding device, comprising: a monitoring device host, a multimodal sensing component, a data analysis module, and a feeding module;

[0007] A multimodal sensing component, comprising several sensors, for monitoring various intestinal information of a patient;

[0008] The monitoring device host, connected to the multimodal sensor assembly, feeding module, and data analysis module, is used to receive and display intestinal information;

[0009] The feeding module is used to feed the patient;

[0010] The data analysis module is used to analyze whether there are any abnormalities in the intestinal information. If so, it analyzes the cause of the abnormality based on the abnormal intestinal information, and triggers the monitoring device host to display and issue an early warning based on the cause of the abnormality, as well as to regulate the feeding module to feed the patient.

[0011] It is also used to comprehensively analyze the patient's intestinal condition based on intestinal information using a constructed spatiotemporal synchronous fusion model, combined with different intestinal information, to generate intestinal condition analysis results, and to trigger the monitoring device host to display and issue warnings based on the intestinal condition analysis results, as well as to regulate the feeding module to feed the patient.

[0012] Furthermore, the intestinal information includes: intestinal imaging and biochemical parameter information;

[0013] The biochemical parameters include one or more of the following: pressure value, three-dimensional motion characteristics, pH value, intestinal gas composition, and concentration of gaseous metabolites.

[0014] Furthermore, the multimodal sensing component includes one or more of the following: a pressure sensor, an accelerometer, a pH sensor, a gas chromatography module, and an image acquisition unit;

[0015] The pressure sensor is used to monitor intestinal pressure values, including: intestinal static pressure, peristaltic contraction pressure, and feeding tube resistance pressure.

[0016] The accelerometer is used to monitor the mechanical vibration of the intestinal wall and obtain the three-dimensional motion characteristics of intestinal peristalsis; wherein the three-dimensional motion characteristics include: frequency, amplitude and propagation direction;

[0017] The pH sensor is used to monitor the pH value of intestinal contents and locate different segments of the intestine;

[0018] The gas chromatography module is used to monitor and analyze the components of intestinal gas and quantify the concentration of gaseous metabolites of intestinal flora; wherein the components of intestinal gas include: O2, CO2, H2 and CH4;

[0019] An image acquisition unit is used to acquire intestinal images in real time; the image acquisition unit includes: a camera and an illumination device; the illumination device is used to provide a light source, and the camera is used to acquire intestinal images.

[0020] Furthermore, the data analysis module is used to analyze whether there are any abnormalities in the intestinal information, analyze the cause of the abnormality based on the abnormal intestinal information, and trigger the monitoring device host to display and issue an early warning based on the cause of the abnormality, as well as to regulate the feeding module to feed the patient, including:

[0021] The system analyzes whether the pressure value falls within the preset normal intestinal lumen pressure range or fluctuates. If not, it determines that the pressure value is abnormal and triggers the monitoring device host to display and issue an alert.

[0022] If the pressure value is not within the preset normal intestinal lumen pressure range, and the pressure value rises above the preset high intestinal lumen pressure threshold, the abnormality is determined to be caused by intestinal obstruction, intestinal stenosis, or feeding tube blockage, triggering the monitoring device host to control the feeding module to stop feeding the patient and perform flushing.

[0023] If the pressure value is not within the preset normal intestinal pressure range, and the pressure value drops below the preset low intestinal pressure threshold, the abnormality is determined to be caused by disordered contractions due to intestinal spasm or inflammation, triggering the monitoring device host to control the feeding module to stop feeding the patient and perform flushing.

[0024] If the pressure value does not fluctuate, the abnormality is determined to be due to intestinal paralysis or severe intestinal motility deficiency, triggering the monitoring device host to adjust the feeding module to reduce the feeding speed of the patient;

[0025] Extract the dominant frequency component from the three-dimensional motion features and analyze whether the dominant frequency component belongs to the preset normal intestinal lumen frequency range. If not, it is determined that there is an abnormality in the three-dimensional motion features.

[0026] If the dominant frequency component does not fall within the preset normal intestinal lumen frequency range and the dominant frequency component is higher than the preset high intestinal lumen frequency threshold, the abnormality is determined to be due to irritable bowel syndrome or hyperperistalsis caused by infectious diarrhea, triggering the monitoring device host to adjust the feeding module to reduce the feeding speed of the patient.

[0027] If the dominant frequency component is not within the preset normal intestinal lumen frequency range and the dominant frequency component is lower than the preset low intestinal lumen frequency threshold, the abnormality is caused by intestinal motility failure, which triggers the monitoring device host to adjust the feeding module to reduce the feeding speed of the patient.

[0028] Analyze whether the propagation direction in the three-dimensional motion characteristics is unidirectional. If not, it is determined that there is an abnormality in the three-dimensional motion characteristics, and the cause of the abnormality is mechanical movement disorder such as intussusception.

[0029] Analyze whether the pH value is within the preset normal intestinal pH range and whether the rate of decrease is greater than the preset rate of decrease or whether there is a change. If not, the pH value is determined to be abnormal.

[0030] If the pH value is not within the preset normal intestinal pH range and the pH value drops to the first preset low intestinal pH threshold, the abnormality is determined to be gastric acid reflux into the intestine, triggering the monitoring device host to control the feeding module to activate the anti-reflux mode;

[0031] If the pH value is not within the preset normal intestinal pH range, and the pH value is greater than the first preset low intestinal pH threshold but less than the second preset low intestinal pH threshold, then the abnormality is determined to be caused by a proximal small bowel gastrinoma or a gastric fistula.

[0032] If the pH value does not change, the abnormality is determined to be due to impaired digestive fluid secretion.

[0033] Analyze whether the concentration of gaseous metabolites meets the preset concentration range or the ratio meets the preset ratio range. If not, it is determined that there is an abnormality in the composition of intestinal gas.

[0034] If the H2 concentration is not within the preset concentration range, the abnormality is determined to be due to small intestinal bacterial overgrowth or lactose intolerance. It is recommended to switch to a lactose-free formula and trigger the monitoring device host to regulate the feeding module to perform antibiotic flushing.

[0035] If the CH4 concentration is not within the preset concentration range, the abnormality is determined to be due to excessive proliferation of methanogens associated with chronic constipation.

[0036] If the H2 / CH4 ratio is not within the preset range, the cause of the abnormality is determined to be an abnormal fermentation type.

[0037] Furthermore, the data analysis module is used to analyze whether there are abnormalities in intestinal information, analyze the cause of the abnormality based on the abnormal intestinal information, and trigger the monitoring device host to display and issue an early warning based on the cause of the abnormality, as well as to regulate the feeding module to feed the patient, and also includes:

[0038] AI image analysis of intestinal images is used to identify whether there are mucosal damage or bleeding points. If so, it is determined that there is an abnormality in the intestine, and the cause of the abnormality is the presence of mucosal damage or bleeding points. It is recommended to increase anti-inflammatory nutritional components.

[0039] If the moving speed of the image acquisition unit is less than the preset moving speed, it is determined that there is an abnormality in the intestine. The cause of the abnormality is mechanical obstruction or severe intestinal paralysis. This triggers the monitoring device host to adjust the feeding module to reduce the feeding speed of the patient.

[0040] Furthermore, the spatiotemporal synchronization fusion model includes: a data preprocessing layer, a feature extraction layer, and a multimodal fusion model;

[0041] The data preprocessing layer is used to perform time alignment and spatial matching of the data collected by each sensor.

[0042] The feature extraction layer is used to extract features from the intestinal information collected by each sensor and output by the data preprocessing layer.

[0043] A multimodal fusion model is used to fuse and analyze the features extracted by the feature extraction layer, and to analyze the intestinal condition.

[0044] Furthermore, the time alignment employs dynamic time warping and Kalman filtering to eliminate differences in the sampling frequencies of each sensor.

[0045]

[0046] in For the optimal time point after alignment, x i (t) is the data vector of the i-th sensor at time t, y k The data vector corresponding to the base timestamp k;

[0047] Spatial registration involves establishing a three-dimensional coordinate system for the intestine based on the SLAM localization of the image acquisition unit, and mapping biochemical parameter information to the anatomical location.

[0048] P s =T e ·P se

[0049] Where P se Let T be the position vector of the sensor in its own coordinate system. e P is the coordinate transformation matrix of the image acquisition unit. s This represents the position of the sensor data in the global gut coordinate system.

[0050] Furthermore, the multimodal fusion model includes: a gut motility assessment model;

[0051] A gut motility assessment model is used to assess gut motility based on features extracted from data from pressure sensors and accelerometers, generating gut motility assessment results, including:

[0052] Based on features extracted from data from pressure sensors and accelerometers, the gut motility index (GMI) is calculated using a bidirectional LSTM network.

[0053]

[0054] Where P t A is the feature vector extracted from the pressure sensor data at time step t. t W is the feature vector extracted from the accelerometer data at time step t. lstm Here is the weight matrix of the LSTM network, b is the bias term, and σ is the Sigmoid activation function, which maps the output to the [0,1] interval. The larger the GMI value, the stronger the intestinal motility.

[0055] Based on features extracted from data from pressure sensors and accelerometers, the intestinal motility state (State) is analyzed.

[0056]

[0057] Where f a P represents the dominant frequency of the three-dimensional motion characteristics. pre The average pressure value is the value of the miniature fiber optic pressure sensor. Waveform entropy is an indicator of the complexity of the pressure waveform; the lower the entropy value, the more regular the waveform.

[0058] GMI and State were used as the results of intestinal motility assessment.

[0059] Furthermore, the multimodal fusion model also includes: an absorption function evaluation model;

[0060] An absorption function assessment model is used to evaluate intestinal absorption function based on pH and the concentration of gaseous metabolites, generating intestinal absorption function assessment results, including:

[0061] Absorption efficiency was used as the result of intestinal absorption function assessment.

[0062]

[0063] Where Δ[H2] is the change in hydrogen concentration per unit time, ΔPH is the change in pH value within the same time period, and η is the absorption efficiency coefficient, with a lower value indicating absorption obstacles.

[0064] Furthermore, the multimodal fusion model also includes: a complication early warning model;

[0065] A complication early warning model is used to assess complications based on intestinal imaging and biochemical parameters, and generate complication assessment results.

[0066] R = α·GMI + β·η + γ·I

[0067] Where α, β, and γ are weighting coefficients, and I is the probability of image anomalies, obtained through CNN;

[0068] The results of intestinal motility assessment, intestinal absorption function assessment, and complication assessment were used as the results of the intestinal condition analysis.

[0069] The beneficial effects of this solution are as follows: This solution sets up a feeding module for feeding and connects to the monitoring equipment host for easy control of feeding. At the same time, it sets up a multimodal sensing component and a data analysis module to collect gastrointestinal data from multiple aspects and analyze the data through the data analysis module to obtain gastrointestinal status, so as to automatically adjust feeding, ensure patient safety, and accelerate the recovery speed.

[0070] The intestinal information monitored by the multimodal sensing components is directly displayed on the main unit of the monitoring device, making it convenient for medical staff to view the raw data directly. The multimodal sensing components include, but are not limited to: miniature fiber optic pressure sensors, triaxial MEMS accelerometers, electrochemical pH sensors, miniature gas chromatography modules, and image acquisition units, thereby collecting intestinal images and biochemical parameter information (pressure value, three-dimensional motion characteristics, pH value, intestinal gas composition, and concentration of gas metabolites). The data analysis module analyzes and deeply mines the collected intestinal information to obtain the patient's intestinal status and adjust feeding.

[0071] Compared to existing technologies, the data analysis module of this solution does not perform analysis on a single indicator range. Instead, it analyzes the data collected by each sensor individually to identify anomalies and, based on the specific data, identifies possible causes of the anomalies to assist medical personnel in diagnosis and treatment. Simultaneously, it combines data from multiple sensors for comprehensive analysis. Because various data in the gut are interconnected, certain anomalies can affect changes in multiple data points, rather than just a single one. Therefore, comprehensive analysis yields more accurate results and improves the accuracy of anomaly identification.

[0072] In summary, this solution enables comprehensive and precise gastrointestinal monitoring of patients with indwelling jejunal feeding tubes, and allows for data mining and analysis to automatically adjust feeding, ensuring patient safety and accelerating recovery. Attached Figure Description

[0073] Figure 1 This is a logic block diagram of an embodiment of an AI-controlled intelligent visual multifunctional fully automatic enteral feeding device according to the present invention. Detailed Implementation

[0074] The following detailed description illustrates the specific implementation method:

[0075] Example 1

[0076] This embodiment is basically as shown in the appendix. Figure 1 As shown: An AI-controlled intelligent visual multifunctional fully automatic intestinal feeding device includes: a monitoring device host, a multimodal sensing component, a data analysis module, and a feeding module;

[0077] A multimodal sensing component, comprising several sensors, for monitoring various intestinal information of a patient; wherein the intestinal information includes, but is not limited to, intestinal imaging and biochemical parameter information;

[0078] The biochemical parameters include one or more of the following: pressure value, three-dimensional motion characteristics, pH value, intestinal gas composition, and concentration of gaseous metabolites.

[0079] The monitoring device host is connected to the multimodal sensor assembly, the feeding module, and the data analysis module to receive and display intestinal information; in this embodiment, the monitoring device host is equipped with a color LCD screen for display.

[0080] The feeding module is used to feed the patient; in this embodiment, the feeding module uses a jejunal feeding tube.

[0081] The data analysis module is used to analyze whether there are any abnormalities in the intestinal information. If so, it analyzes the cause of the abnormality based on the abnormal intestinal information, and triggers the monitoring device host to display and issue an early warning based on the cause of the abnormality, as well as to regulate the feeding module to feed the patient.

[0082] It is also used to comprehensively analyze the patient's intestinal condition based on intestinal information using a constructed spatiotemporal synchronous fusion model, combined with different intestinal information, to generate intestinal condition analysis results, and to trigger the monitoring device host to display and issue warnings based on the intestinal condition analysis results, as well as to regulate the feeding module to feed the patient.

[0083] The aforementioned multimodal sensing components include one or more of the following: a pressure sensor, an accelerometer, a pH sensor, a gas chromatography module, and an image acquisition unit; in this embodiment, a miniature fiber optic pressure sensor, a triaxial MEMS accelerometer, an electrochemical pH sensor, and a miniature gas chromatography module are used.

[0084] The miniature fiber optic pressure sensor is used to monitor intestinal pressure values, including: intestinal static pressure, peristaltic contraction pressure, and feeding tube resistance pressure.

[0085] A triaxial MEMS accelerometer is used to monitor the mechanical vibration of the intestinal wall and obtain the three-dimensional motion characteristics of intestinal peristalsis; the three-dimensional motion characteristics include: frequency, amplitude and propagation direction;

[0086] An electrochemical pH sensor is used to monitor the pH value (i.e., acidity / alkalinity) of intestinal contents and to locate different sections of the intestine.

[0087] The miniature gas chromatography module is used to monitor and analyze the composition of intestinal gases and quantify the concentration of gaseous metabolites of intestinal flora; the intestinal gas components include: O2 / CO2 / H2 / CH4;

[0088] An image acquisition unit is used to acquire intestinal images in real time. The image acquisition unit includes a camera and an illumination device. The illumination device is used to provide a light source, and the camera is used to acquire intestinal images. In this embodiment, the camera is a biodegradable capsule endoscope, and the illumination device is a dual-wavelength laser illumination device. The battery life is required to cover the entire small intestine examination cycle. Both are inside the capsule.

[0089] In addition, in this embodiment, the miniature fiber optic pressure sensor, triaxial MEMS accelerometer, electrochemical pH sensor and miniature gas chromatography module are disposed on the outer wall of the jejunal feeding tube. In other embodiments, one or more sensors may be disposed inside the capsule.

[0090] The above analysis checks for abnormalities in intestinal information. Based on the abnormal intestinal information, the cause of the abnormality is analyzed. Based on the cause of the abnormality, the monitoring device host is triggered to display and issue an early warning, and the feeding module is adjusted to feed the patient. The specific process is as follows:

[0091] The system analyzes whether the pressure value falls within the preset normal intestinal lumen pressure range or fluctuates. If not, it determines that the pressure value is abnormal and triggers the monitoring device host to display and issue an early warning. In this embodiment, the preset normal intestinal lumen pressure range is 10-30 mmHg.

[0092] If the pressure value is not within the preset normal intestinal lumen pressure range, and the pressure value rises above the preset high intestinal lumen pressure threshold, the abnormality is determined to be caused by intestinal obstruction, intestinal stenosis, or feeding tube blockage. This triggers the monitoring device host to control the feeding module to stop feeding the patient and perform flushing. In this embodiment, the preset high intestinal lumen pressure threshold is 50 mmHg.

[0093] If the pressure value is not within the preset normal intestinal lumen pressure range, and the pressure value decreases below the preset low intestinal lumen pressure threshold, the abnormality is determined to be caused by disordered contractions due to intestinal spasm or inflammation, triggering the monitoring device host to control the feeding module to stop feeding the patient and perform flushing; in this embodiment, the preset low intestinal lumen pressure threshold is 5 mmHg;

[0094] If the pressure value does not fluctuate, the abnormality is determined to be due to intestinal paralysis (common after surgery) or severe intestinal motility deficiency, which triggers the monitoring device host to adjust the feeding module to reduce the feeding speed of the patient.

[0095] The dominant frequency component of the three-dimensional motion features is extracted, and it is analyzed whether the dominant frequency component belongs to the preset normal intestinal lumen frequency range. If not, it is determined that there is an abnormality in the three-dimensional motion features. In this embodiment, the preset normal intestinal lumen frequency range is 0.2-0.5Hz.

[0096] If the dominant frequency component does not fall within the preset normal intestinal lumen frequency range and is higher than the preset intestinal lumen high frequency threshold, the abnormality is determined to be due to irritable bowel syndrome or hyperperistalsis caused by infectious diarrhea. This triggers the monitoring device host to adjust the feeding module to reduce the feeding speed of the patient. In this embodiment, the preset intestinal lumen high frequency threshold is 1Hz.

[0097] If the dominant frequency component does not fall within the preset normal intestinal lumen frequency range and is lower than the preset low intestinal lumen frequency threshold, the abnormality is due to intestinal motility failure, triggering the monitoring device host to adjust the feeding module to reduce the feeding speed of the patient; in this embodiment, the preset low intestinal lumen frequency threshold is 0.1.

[0098] Analyze whether the propagation direction in the three-dimensional motion characteristics is unidirectional. If not, it is determined that there is an abnormality in the three-dimensional motion characteristics, and the cause of the abnormality is mechanical movement disorder such as intussusception.

[0099] The pH value is analyzed to determine whether it falls within the preset normal intestinal pH range and whether the rate of decrease is greater than the preset rate of decrease or whether there is a change. If not, the pH value is determined to be abnormal. In this embodiment, the preset normal intestinal pH range is 6.5-7.5, and the preset rate of decrease is 0.3 / min.

[0100] If the pH value is not within the preset normal intestinal pH range and drops to the first preset low intestinal pH threshold, the abnormality is determined to be gastric acid reflux into the intestine, triggering the monitoring device host to control the feeding module to activate the anti-reflux mode and start the proton pump inhibitor; in this embodiment, the first preset low intestinal pH threshold is 4.5;

[0101] If the pH value is not within the preset normal intestinal pH range, and the pH value is greater than the first preset low intestinal pH threshold but less than the second preset low intestinal pH threshold, then the abnormality is determined to be caused by a proximal small bowel gastrinoma or gastric fistula; in this embodiment, the second preset low intestinal pH threshold is 6.0.

[0102] If the pH value does not change, the abnormality is determined to be due to impaired digestive fluid secretion.

[0103] The concentration of gaseous metabolites is analyzed to see if it meets the preset concentration range or the ratio meets the preset ratio range. If not, it is determined that there is an abnormality in the intestinal gas composition. In this embodiment, the preset concentration range is H2 concentration less than 100ppm and CH4 concentration less than 50ppm; the preset ratio range is H2 / CH4 less than 4.

[0104] If the H2 concentration is not within the preset concentration range, the abnormality is determined to be due to small intestinal bacterial overgrowth (SIBO) or lactose intolerance. It is recommended to switch to a lactose-free formula and trigger the monitoring device host to regulate the feeding module to perform antibiotic flushing.

[0105] If the CH4 concentration is not within the preset concentration range, the abnormality is determined to be due to excessive proliferation of methanogens associated with chronic constipation.

[0106] If the H2 / CH4 ratio is not within the preset range, the cause of the abnormality is determined to be an abnormal fermentation type.

[0107] AI image analysis of intestinal images is used to identify whether there are mucosal damages or bleeding points. If so, it is determined that there is an abnormality in the intestine, and the cause of the abnormality is the presence of mucosal damage or bleeding points. It is recommended to increase anti-inflammatory nutritional components. In this embodiment, CNN recognition is used for AI image analysis.

[0108] If the moving speed of the image acquisition unit is less than the preset moving speed, it is determined that there is an abnormality in the intestine. The cause of the abnormality is mechanical obstruction or severe intestinal paralysis. This triggers the monitoring device host to adjust the feeding module to reduce the feeding speed of the patient.

[0109] The aforementioned data analysis module is also used to comprehensively analyze the patient's intestinal condition based on intestinal information using a constructed spatiotemporal synchronous fusion model, combined with different intestinal information, generating intestinal condition analysis results. Based on the intestinal condition analysis results, it triggers the monitoring device host to display and issue warnings, and regulates the feeding module to feed the patient. The specific process is as follows:

[0110] The spatiotemporal synchronization fusion model includes: a data preprocessing layer, a feature extraction layer, and a multimodal fusion model;

[0111] The data preprocessing layer is used to perform time alignment and spatial matching of the data collected by each sensor.

[0112] The time alignment employs Dynamic Time Warping (DTW) and Kalman filtering to eliminate differences in the sampling frequencies of the various sensors.

[0113]

[0114] in For the optimal time point after alignment, x i (t) is the data vector of the i-th sensor at time t, y k This represents the data vector corresponding to the reference timestamp k. It aligns time-series data from different sensors to resolve timing inconsistencies caused by different sampling frequencies, such as high-frequency sampling from pressure sensors versus low-frequency sampling from pH sensors. Kalman filtering eliminates sensor noise, improves the signal-to-noise ratio of the data, and provides a more stable input for time alignment.

[0115] Spatial registration involves establishing a three-dimensional coordinate system for the intestine based on the SLAM (simultaneous localization and mapping) positioning of the image acquisition unit, and mapping biochemical parameter information to anatomical locations.

[0116] P s =T e ·P se

[0117] Where P se Let T be the position vector of the sensor in its own coordinate system. e P is the coordinate transformation matrix of the image acquisition unit. s This represents the position of the sensor data in the global gut coordinate system.

[0118] The feature extraction layer is used to extract features from the intestinal information collected by each sensor output by the data preprocessing layer. In this embodiment, wavelet transform is used to extract the peristaltic wave peak frequency, amplitude, and propagation speed from the data collected by the miniature fiber optic pressure sensor.

[0119] The data collected by the triaxial MEMS accelerometer were processed using fast Fourier transform to extract the dominant frequency and waveform entropy of intestinal wall vibration.

[0120] Piecewise linear regression was used to extract the slope of acidity change and steady-state range from the data collected by the electrochemical pH sensor.

[0121] Principal component analysis was used to extract gas concentration gradients and metabolite ratios from the data collected by the micro gas chromatography module.

[0122] The data acquired by the image acquisition unit is processed using a convolutional neural network to extract mucosal texture, blood vessel density, and motion artifacts.

[0123] A multimodal fusion model is used to fuse and analyze the features extracted by the feature extraction layer, and to analyze the intestinal condition.

[0124] The multimodal fusion model includes: intestinal motility assessment model, absorption function assessment model, and complication early warning model;

[0125] The intestinal motility assessment model is used to assess intestinal motility based on features extracted from data from a miniature fiber optic pressure sensor and a triaxial MEMS accelerometer, and to generate intestinal motility assessment results.

[0126] Specifically, the intestinal motility index GMI is calculated using a bidirectional LSTM network based on features extracted from data from a miniature fiber optic pressure sensor and a triaxial MEMS accelerometer.

[0127]

[0128] Where P t A is a feature vector extracted from the data of a miniature fiber optic pressure sensor at time step t. t W is the feature vector extracted from the data of the triaxial MEMS accelerometer at time step t. lstm Here is the weight matrix of the LSTM network, b is the bias term, and σ is the Sigmoid activation function, which maps the output to the [0,1] interval. The larger the GMI value, the stronger the intestinal motility.

[0129] Based on features extracted from data from a miniature fiber optic pressure sensor and a triaxial MEMS accelerometer, the intestinal motility state (State) was analyzed.

[0130]

[0131] Where f a P represents the dominant frequency of the three-dimensional motion characteristics. pre The average pressure value is the value of the miniature fiber optic pressure sensor. Waveform entropy is an indicator of the complexity of the pressure waveform; the lower the entropy value, the more regular the waveform.

[0132] GMI and State were used as the results of intestinal motility assessment.

[0133] An absorption function assessment model is used to assess intestinal absorption function based on pH and concentration of gaseous metabolites, and to generate intestinal absorption function assessment results.

[0134] Specifically, absorption efficiency is used as the assessment result of intestinal absorption function:

[0135]

[0136] Where Δ[H2] is the change in hydrogen concentration per unit time (reflecting the metabolic activity of the microbial community), ΔPH is the change in pH value within the same time period (reflecting the secretion of digestive juices), and η is the absorption efficiency coefficient, the lower the value, the more likely it is to indicate absorption impairment.

[0137] A complication early warning model is used to assess complications based on intestinal imaging and biochemical parameters, and generate complication assessment results.

[0138] R = α·GMI + β·η + γ·I

[0139] Where α, β, and γ are weight coefficients, and I is the probability of image anomalies, obtained through CNN.

[0140] The results of intestinal motility assessment, intestinal absorption function assessment, and complication assessment were used as the analysis results of intestinal condition.

[0141] In addition, in this embodiment, the monitoring device host is triggered to display and issue an early warning based on the cause of the anomaly, and a yellow warning is used;

[0142] Based on the intestinal condition analysis results, including the intestinal motility assessment results and the intestinal absorption function assessment results, the monitoring equipment host is triggered to display and issue an early warning, using an orange warning;

[0143] Based on the complication assessment results in the intestinal condition analysis, the monitoring device host is triggered to display and issue an early warning, using a red warning;

[0144] The three warning modes are set based on the accuracy of anomaly detection. The yellow warning is when a single sensor detects an anomaly, the orange warning is when two sensors work together to detect an anomaly, and the red warning is when multiple sensors detect anomalies, resulting in higher accuracy. The multi-level warning system makes it easier to distinguish the warning level and take timely action.

[0145] In other embodiments, corresponding handling strategies are set for the three different warnings: a yellow warning reduces the speed of the feeding module and rechecks for abnormalities; an orange warning stops the feeding module from feeding and rinses the module, and rechecks for abnormalities; and a red warning stops the feeding module from feeding and triggers an audible and visual alarm.

[0146] This solution includes a feeding module for feeding and connects to a monitoring device host for easy control of feeding. It also includes multimodal sensing components and a data analysis module to collect gastrointestinal data from various aspects and analyze the data to obtain gastrointestinal information, thereby automatically adjusting feeding to ensure patient safety and accelerate recovery.

[0147] The intestinal information monitored by the multimodal sensing components is directly displayed on the main unit of the monitoring device, making it convenient for medical staff to view the raw data directly. The multimodal sensing components include, but are not limited to: miniature fiber optic pressure sensors, triaxial MEMS accelerometers, electrochemical pH sensors, miniature gas chromatography modules, and image acquisition units, thereby collecting intestinal images and biochemical parameter information (pressure value, three-dimensional motion characteristics, pH value, intestinal gas composition, and concentration of gas metabolites). The data analysis module analyzes and deeply mines the collected intestinal information to obtain the patient's intestinal status and adjust feeding.

[0148] The data analysis module does not perform single-indicator range analysis, but rather conducts anomaly analysis on the data collected by each sensor one by one, and analyzes the possible causes of anomalies based on the specific data to assist medical staff in diagnosis and treatment. At the same time, it combines data from multiple sensors for comprehensive analysis, because there are interrelationships among various data in the intestine, and certain anomalies will affect the changes of multiple data rather than a single data change. Therefore, comprehensive analysis is conducted to obtain more accurate analysis results and improve the accuracy of anomaly identification.

[0149] This comprehensive analysis process, encompassing data alignment, feature extraction, multimodal fusion, and clinical decision-making, utilizes mathematical modeling to accurately align spatiotemporal data, eliminating errors caused by hardware differences; quantifying physiological characteristics by transforming intestinal function into calculable indicators; and providing dynamic risk assessment by integrating multi-dimensional information in real time to offer clinical early warnings. This approach is tailored to clinical needs, ensuring its practicality.

[0150] Compared to traditional single-data-range judgment, this solution provides a more comprehensive analysis and evaluation, more accurate abnormal monitoring, full utilization of data, automatic adjustment of feeding, ensures patient safety, and accelerates recovery.

[0151] Example 2

[0152] This embodiment is basically the same as the above embodiment, except that it also includes: the main unit of the monitoring device is provided with several slots for connecting different functional components to realize different functional combinations; in this embodiment, there are different slots at the rear of the monitoring device, and different functional components can be inserted into different slots to work, which makes it convenient for users to arrange and combine different functional components. The functional components can be set according to the characteristics of the patient's main disease, such as lung disease function, kidney disease function, brain disease function, heart disease function, etc. These disease functions can help doctors make automatic selections, such as adding blood oxygen monitoring function components, heart rate monitoring function components, etc., so as to make the monitoring more comprehensive and systematic, with real-time image and waveform analysis of the whole process, and scientific and precise feeding based on the collected data. It is a device that is very much needed in intensive care units and geriatric nursing hospitals.

[0153] The above descriptions are merely embodiments of the present invention. Commonly known structures and characteristics are not described in detail here. Those skilled in the art are aware of all common technical knowledge in the field prior to the application date or priority date, are aware of all existing technologies in that field, and have the ability to apply conventional experimental methods prior to that date. Those skilled in the art can, under the guidance of this application, improve and implement this solution in combination with their own capabilities. Some typical known structures or methods should not be obstacles for those skilled in the art to implement this application. It should be noted that those skilled in the art can make several modifications and improvements without departing from the structure of the present invention. These should also be considered within the scope of protection of the present invention, and will not affect the effectiveness of the implementation of the present invention or the practicality of the patent. The scope of protection claimed in this application should be determined by the content of its claims, and the specific embodiments described in the specification can be used to interpret the content of the claims.

Claims

1. An AI-controlled intelligent visual multifunctional fully automatic enteral feeding device, characterized in that: include: Monitoring equipment host, multimodal sensing components, data analysis module, and feeding module; A multimodal sensing component, comprising several sensors, for monitoring various intestinal information of a patient; The monitoring device host, connected to the multimodal sensor assembly, feeding module, and data analysis module, is used to receive and display intestinal information; The feeding module is used to feed the patient; The data analysis module is used to analyze whether there are any abnormalities in the intestinal information. If so, it analyzes the cause of the abnormality based on the abnormal intestinal information, and triggers the monitoring device host to display and issue an early warning based on the cause of the abnormality, as well as to regulate the feeding module to feed the patient. It is also used to comprehensively analyze the patient's intestinal condition based on intestinal information using a constructed spatiotemporal synchronous fusion model, combined with different intestinal information, to generate intestinal condition analysis results, and to trigger the monitoring device host to display and issue warnings based on the intestinal condition analysis results, as well as to regulate the feeding module to feed the patient.

2. The AI-controlled intelligent visual multifunctional fully automatic enteral feeding device according to claim 1, characterized in that: The intestinal information includes: intestinal imaging and biochemical parameter information; The biochemical parameters include one or more of the following: pressure value, three-dimensional motion characteristics, pH value, intestinal gas composition, and concentration of gaseous metabolites.

3. The AI-controlled intelligent visual multifunctional fully automatic enteral feeding device according to claim 2, characterized in that: The multimodal sensing component includes one or more of the following: a pressure sensor, an accelerometer, a pH sensor, a gas chromatography module, and an image acquisition unit. The pressure sensor is used to monitor the pressure values ​​in the intestine; the pressure values ​​include: intestinal static pressure, peristaltic contraction pressure, and feeding tube resistance pressure; The accelerometer is used to monitor the mechanical vibration of the intestinal wall and obtain the three-dimensional motion characteristics of intestinal peristalsis; wherein the three-dimensional motion characteristics include: frequency, amplitude and propagation direction; The pH sensor is used to monitor the pH value of intestinal contents and locate different segments of the intestine; The gas chromatography module is used to monitor and analyze the components of intestinal gas and quantify the concentration of gaseous metabolites of intestinal flora; wherein the components of intestinal gas include: O2, CO2, H2 and CH4; An image acquisition unit is used to acquire intestinal images in real time; the image acquisition unit includes: a camera and an illumination device; the illumination device is used to provide a light source, and the camera is used to acquire intestinal images.

4. The AI-controlled intelligent visual multifunctional fully automatic enteral feeding device according to claim 3, characterized in that: The data analysis module is used to analyze whether there are any abnormalities in intestinal information, analyze the causes of abnormalities based on the abnormal intestinal information, and trigger the monitoring device host to display and issue warnings based on the causes of abnormalities, as well as to regulate the feeding module to feed the patient, including: The system analyzes whether the pressure value falls within the preset normal intestinal lumen pressure range or fluctuates. If not, it determines that the pressure value is abnormal and triggers the monitoring device host to display and issue an early warning. If the pressure value is not within the preset normal intestinal lumen pressure range, and the pressure value rises above the preset high intestinal lumen pressure threshold, the abnormality is determined to be caused by intestinal obstruction, intestinal stenosis, or feeding tube blockage, triggering the monitoring device host to control the feeding module to stop feeding the patient and perform flushing. If the pressure value is not within the preset normal intestinal pressure range, and the pressure value drops below the preset low intestinal pressure threshold, the abnormality is determined to be caused by disordered contractions due to intestinal spasm or inflammation, triggering the monitoring device host to control the feeding module to stop feeding the patient and perform flushing. If the pressure value does not fluctuate, the abnormality is determined to be due to intestinal paralysis or severe intestinal motility deficiency, triggering the monitoring device host to adjust the feeding module to reduce the feeding speed of the patient; Extract the dominant frequency component from the three-dimensional motion features and analyze whether the dominant frequency component belongs to the preset normal intestinal lumen frequency range. If not, it is determined that there is an abnormality in the three-dimensional motion features. If the dominant frequency component does not fall within the preset normal intestinal lumen frequency range and the dominant frequency component is higher than the preset high intestinal lumen frequency threshold, the abnormality is determined to be due to irritable bowel syndrome or hyperperistalsis caused by infectious diarrhea, triggering the monitoring device host to adjust the feeding module to reduce the feeding speed of the patient. If the dominant frequency component is not within the preset normal intestinal lumen frequency range and the dominant frequency component is lower than the preset low intestinal lumen frequency threshold, the abnormality is caused by intestinal motility failure, which triggers the monitoring device host to adjust the feeding module to reduce the feeding speed of the patient. Analyze whether the propagation direction in the three-dimensional motion characteristics is unidirectional. If not, it is determined that there is an abnormality in the three-dimensional motion characteristics, and the cause of the abnormality is mechanical movement disorder such as intussusception. Analyze whether the pH value is within the preset normal intestinal pH range and whether the rate of decrease is greater than the preset rate of decrease or whether there is a change. If not, the pH value is determined to be abnormal. If the pH value is not within the preset normal intestinal pH range and the pH value drops to the first preset low intestinal pH threshold, the abnormality is determined to be gastric acid reflux into the intestine, triggering the monitoring device host to control the feeding module to activate the anti-reflux mode; If the pH value is not within the preset normal intestinal pH range, and the pH value is greater than the first preset low intestinal pH threshold but less than the second preset low intestinal pH threshold, then the abnormality is determined to be caused by a proximal small bowel gastrinoma or gastric fistula. If the pH value does not change, the abnormality is determined to be due to impaired digestive fluid secretion. Analyze whether the concentration of gaseous metabolites meets the preset concentration range or the ratio meets the preset ratio range. If not, it is determined that there is an abnormality in the composition of intestinal gas. If the H2 concentration is not within the preset concentration range, the abnormality is determined to be due to small intestinal bacterial overgrowth or lactose intolerance. It is recommended to switch to a lactose-free formula and trigger the monitoring device host to regulate the feeding module to perform antibiotic flushing. If the CH4 concentration is not within the preset concentration range, the abnormality is determined to be due to excessive proliferation of methanogens associated with chronic constipation. If the H2 / CH4 ratio is not within the preset range, the cause of the abnormality is determined to be an abnormal fermentation type.

5. The AI-controlled intelligent visual multifunctional fully automatic enteral feeding device according to claim 4, characterized in that: The data analysis module is used to analyze whether there are abnormalities in intestinal information, analyze the cause of the abnormality based on the abnormal intestinal information, and trigger the monitoring device host to display and issue an early warning based on the cause of the abnormality, as well as to regulate the feeding module to feed the patient, and also includes: AI image analysis of intestinal images is used to identify whether there are mucosal damage or bleeding points. If so, it is determined that there is an abnormality in the intestine, and the cause of the abnormality is the presence of mucosal damage or bleeding points. It is recommended to increase anti-inflammatory nutritional components. If the moving speed of the image acquisition unit is less than the preset moving speed, it is determined that there is an abnormality in the intestine. The cause of the abnormality is mechanical obstruction or severe intestinal paralysis. This triggers the monitoring device host to adjust the feeding module to reduce the feeding speed of the patient.

6. The AI-controlled intelligent visual multifunctional fully automatic enteral feeding device according to claim 5, characterized in that: The spatiotemporal synchronization fusion model includes: a data preprocessing layer, a feature extraction layer, and a multimodal fusion model; The data preprocessing layer is used to perform time alignment and spatial matching of the data collected by each sensor. The feature extraction layer is used to extract features from the intestinal information collected by each sensor and output by the data preprocessing layer. A multimodal fusion model is used to fuse and analyze the features extracted by the feature extraction layer, and to analyze the intestinal condition.

7. The AI-controlled intelligent visual multifunctional fully automatic enteral feeding device according to claim 6, characterized in that: The time alignment is achieved by using dynamic time warping and Kalman filtering to eliminate differences in the sampling frequencies of the various sensors. in For the optimal time point after alignment, x i (t) is the data vector of the i-th sensor at time t, y k The data vector corresponding to the base timestamp k; Spatial registration involves establishing a three-dimensional coordinate system for the intestine based on the SLAM localization of the image acquisition unit, and mapping biochemical parameter information to the anatomical location. P s =T e ·P se Where P se Let T be the position vector of the sensor in its own coordinate system. e P is the coordinate transformation matrix of the image acquisition unit. s This represents the position of the sensor data in the global gut coordinate system.

8. The AI-controlled intelligent visual multifunctional fully automatic enteral feeding device according to claim 7, characterized in that: The multimodal fusion model includes: a gut motility assessment model; A gut motility assessment model is used to assess gut motility based on features extracted from data from pressure sensors and accelerometers, generating gut motility assessment results, including: Based on features extracted from data from pressure sensors and accelerometers, the gut motility index (GMI) is calculated using a bidirectional LSTM network. Where P t A is the feature vector extracted from the pressure sensor data at time step t. t W is the feature vector extracted from the accelerometer data at time step t. lstm Here is the weight matrix of the LSTM network, b is the bias term, and σ is the Sigmoid activation function, which maps the output to the [0,1] interval. The larger the GMI value, the stronger the intestinal motility. Based on features extracted from data from pressure sensors and accelerometers, the intestinal motility state (State) is analyzed. Where f a P represents the dominant frequency of the three-dimensional motion characteristics. pre The average pressure value is the value of the miniature fiber optic pressure sensor. Waveform entropy is an indicator of the complexity of the pressure waveform; the lower the entropy value, the more regular the waveform. GMI and State were used as the results of intestinal motility assessment.

9. The AI-controlled intelligent visual multifunctional fully automatic enteral feeding device according to claim 8, characterized in that: The multimodal fusion model also includes: an absorption function evaluation model; An absorption function assessment model is used to evaluate intestinal absorption function based on pH and the concentration of gaseous metabolites, generating intestinal absorption function assessment results, including: Absorption efficiency was used as the result of intestinal absorption function assessment. Where Δ[H2] is the change in hydrogen concentration per unit time, ΔPH is the change in pH value within the same time period, and η is the absorption efficiency coefficient, with a lower value indicating absorption obstacles.

10. The AI-controlled intelligent visual multifunctional fully automatic enteral feeding device according to claim 9, characterized in that: The multimodal fusion model also includes: a complication early warning model; A complication early warning model is used to assess complications based on intestinal imaging and biochemical parameters, and generate complication assessment results. R = α·GMI + β·η + γ·I Where α, β, and γ are weighting coefficients, and I is the probability of image abnormality, which is obtained by analyzing intestinal images using CNN. The results of intestinal motility assessment, intestinal absorption function assessment, and complication assessment were used as the results of the intestinal condition analysis.

Images